Traceability model of malicious messages against data sparsity in time-varying networks

Abstract

This paper proposes an innovative traceability model to address the issue of tracing malicious information in time-varying social networks, overcoming the limitations of traditional methods in dealing with dynamic structures and sparse historical data. This study introduces a time-window mechanism, simplifying the dynamic topology into static snapshots, thereby capturing the time-varying characteristics of network topology more accurately. To tackle the challenge of sparse historical data on malicious information propagation, this paper creatively combines the independent cascade model, maximum likelihood estimation, and probabilistic graphs, significantly improving the accuracy of calculating the likelihood of a node being the source of propagation. Furthermore, a new propagation similarity metric is proposed, and a global expectation function within an attention mechanism is introduced to assign weights to the propagation source based on network stability, significantly reducing the complexity of traditional traceability algorithms. Experimental results demonstrate that the proposed model exhibits excellent performance in practical applications, accurately and timely identifying malicious information sources even with limited public dataset availability. This study not only introduces a novel model and methodology but also provides new insights and directions for future research on malicious information traceability in dynamic social networks.